Hi Laine,
Thanks a lot for your inputs. We will try to download the code from libvirt-0.10.2-main t and build libvirt with the patch for crash issue.
We are using WR kenerl of below version
2.6.34.13-WR4
Thanks
Abhinay
On Wed, Sep 4, 2013 at 1:36 PM, Laine Stump <laine@xxxxxxxxx> wrote:
On 09/04/2013 05:06 AM, arun abhinay wrote:libvirt-0.10.2-19.el6 is libvirt-0.10.2 with *many* bugfix patches
>
> Thanks for your detailed information. As part of fixing this issue we
> have upgraded to libvirt-1.0.5 version and these crashes were
> resolved. But after this upgrade we observed many other issues like
> "cpu-stats" command is not working and vcpu pinning was failing with
> errors. Since we approaching near to our product release date we have
> planned to roll back libvirt-0.10.2 where every thing was working fine
> except for the crash.
>
> After roll back i have tried applying the patches mentioned in comment
> 17 on libvirt-0.10.2 to resolve the crash but this gave me another
> crash whose BT is mentioned below. Hence i thought of upgrading to
> libvirt-0.10.2-19.e16 which i thought would be same level as
> libvirt-0.10.2.
(handpicked from later libvirt releases) applied, as well as some
patches that are specific to RHEL only (and wouldn't make sense for a
non-RHEL distro).
You haven't said what OS you are running (aside from referencing a bug
>
> (gdb) bt
> #0 0x00007f5ab7b89005 in raise () from /lib64/libc.so.6
> #1 0x00007f5ab7b8be40 in abort () from /lib64/libc.so.6
> #2 0x00007f5ab7b82191 in __assert_fail () from /lib64/libc.so.6
> #3 0x00007f5ab8868bb5 in __pthread_mutex_lock_full () from
> /lib64/libpthread.so.0
> #4 0x00007f5aba7904b7 in ?? () from /usr/lib64/libvirt.so.0
> #5 0x00007f5aba7b1172 in ?? () from /usr/lib64/libvirt.so.0
> #6 0x00007f5aba7b3ac1 in ?? () from /usr/lib64/libvirt.so.0
> #7 0x00007f5aba6baab5 in virEventPollRunOnce () from
> /usr/lib64/libvirt.so.0
> #8 0x00007f5aba6b96e5 in virEventRunDefaultImpl () from
> /usr/lib64/libvirt.so.0
> #9 0x000000000040df22 in ?? ()
> #10 0x00007f5aba6ccf36 in ?? () from /usr/lib64/libvirt.so.0
> #11 0x00007f5ab88666ea in start_thread () from /lib64/libpthread.so.0
> #12 0x00007f5ab7c27abd in clone () from /lib64/libc.so.6
>
> I now have only one option to try debug above crash. If you can
> provide me some pointers for above i would be very grateful to you.
>
>
report filed against a build of libvirt made for RHEL6.x). If you're
running RHEL, then you should open a support incident with Red Hat. If
you're running some other distro, you should find the most recent
pre-built package for that distro. If that doesn't help and you have to
build libvirt yourself, then rather than upgrading to some different old
version (1.0.5), you would probably do better to upgrade to the latest
release (1.1.2).
Another possibility I thought of was to checkout the v0.10.2-maint
branch from libvirt git and build that, but it unfortunately appears
that some patches that were backported from upstream master to the RHEL6
build haven't been backported to the upstream v0.10.2-maint branch (and
the series of patches that fix the bug in question are in the "not
backported" category). However, you could start with a checkout of that
branch, then add the patches for this bug - that would at least give
better results than starting with vanilla v0.10.2 release sources and
applying the patches.
-- libvir-list mailing list libvir-list@xxxxxxxxxx https://www.redhat.com/mailman/listinfo/libvir-list